Advanced Energy Industries, Inc. (AEIS) Earnings Call Transcript & Summary

I guess we can get started. Hi. I'm Shane Brett, U.S. semiconductor equipment analyst. Joining me today from Advanced Energy are Steve Kelley, President and CEO; and Paul Oldham, EVP and CFO. Before I start my questions, I'm going to hand it over to Paul for a quick Reg FD disclosure. And for Steve, if you want to give kind of an elevator pitch of what Advanced Energy does after that.

Yes. Thanks a lot, Shane. Thanks for joining us today. Just a reminder that any comments we make today may be subject to certain risk factors. You can find a better discussion of those in our SEC filings. Also a reminder that we are not providing any update to guidance today. Our earnings release was on November 4. So please refer to that as our latest guidance.

Okay. Maybe a quick overview of Advanced Energy. We're headquartered in Denver, Colorado. We're about a 45-year-old company. We got our start providing power solutions for etch chamber makers, so for Applied Materials, Lam and others in that business. So about half of our business today is in semiconductor equipment. We provide these critical subsystems. The other half is what we call system power. So most of that business is for AI data center, but we're also a significant player in industrial and medical products. And in all the markets we play in the high end, over 70% of our revenue comes from sole-source products. So we're on a journey to get that above 80%. And our goal over the next years, as articulated in our Investor Day last November, is to double the size of the company to $3 billion in revenue and about $15 in earnings per share.

Great. As a semiconductor equipment analyst, I've always thought of you guys as a semiconductor equipment company, but I think you're probably a data center company now or -- I just want to kind of talk about the data center business as you're on track to double it this year. Just what's driven this growth? Can you give us kind of a bit of color on what's the drivers and what's kind of the visibility into '26 on data center?

Yes. So when we bought Artisan back in 2019, they were in the data center business, but they were supplying commodity solutions. So we faced the decision roughly 3 to 4 years ago about whether to exit the business because of the low margins or to reorient the business. So we chose to stay in the business and focus our engineering team on higher-end solutions. So trying to tackle the most difficult problems. And so that strategy has worked out well for us. Today, the margins we generate in the data center business are just below corporate average, which is a huge increase over what we used to be able to generate. We're engaged with a select group of hyperscale customers. We're going to grow over 100% this year relative to revenue in 2024. And we forecasted to grow 25% to 30% in '26. And this is solely with our first wave customers, the select group of hyperscalers that we've engaged with. We're also starting discussions with second wave customers, which don't require the same level of engineering intensity. We could modify basically standard solutions to meet their needs. So there's some upside looking forward. We think most of that upside occurs in '27.

I'd say 100% growth is probably a little bit better than going well, but -- yes, it is going well.

We're pleased.

Yes. So I guess there's just so many data center projects going on right now. And I think just within Morgan Stanley researchers seem to consistently be raising numbers. But where are you guys in terms of meeting this data center demand? And how big of a challenge has it been to sort of expand capacity to meet this demand?

Yes. So there's basically 2 areas where we need to invest. One is in development and the second is in production capacity. So development is largely a function of our engineering team. It's also a function of our investment because these higher power products require more expensive labs and development equipment. So we've invested a lot. We've invested much more than we normally do in a year. Actually, the last couple of years, we've overinvested in CapEx, but -- and most of that's gone for data center capacity and development products. So we have kept up with demand. We have expanded capacity in the Philippines and Mexico. And we have a third factory will likely ramp next year for data center in Thailand.

But in terms of this new capacity there, like the payback period isn't a matter of years. It's almost a month.

Yes. So typically, our payback on this new investment is 9 months or shorter. So it's a quick payback. And the equipment is fungible, which means we could use it for other products such as semiconductor, industrial and medical.

Got it. I just want to kind of follow up on that 25% to 30% growth you kind of talked about for 2026. Granted you are a sole-source business, I think the hyperscalers will probably give you a certain amount of visibility. How much does that -- how much does your visibility from these customers extend? And maybe how early are you kind of brought into new projects with these hyperscalers?

Yes. So our visibility, I would say, is roughly 9 months. We get a forecast and sometimes a PO for 9 months. So the visibility is good. We need to be involved with our customers on these designs very early because what we're seeing now is a yearly cadence of new products and a lot of it is tied into the introduction of new GPUs. And so for every new GPU, you typically need a new power solution because each GPU typically uses more power. And so today, we have secured the design wins necessary to support our forecast for '26. So what we're actually working on with customers today are designs for '27 and '28. And so that's really the cadence is we're looking at not M+1, but M+ 2 and beyond with our customers.

Got it. Got it. I guess a little bit more of a longer-term question, which may be interpreted as a short-term question. But how would you think about the longer-term growth prospects of this business? And is there sort of any way you've thought about your attach rate sort of revenue opportunity per gigawatt or like dollar of hyperscaler CapEx that you've sort of thought about?

Yes, it's difficult to identify an attach rate or some growth with the industry because if you think about it, our whole strategy is to be a specialty player. It's to focus where there's unique opportunities for customization where we can bring great technology to market. I think what we could say, though, is that as this data center market evolves and continues to require more power and more capability that our content at each of those levels should continue to increase.

Got it. And would there be a bit of a content story with these kind of second wave of customers that you've identified?

Yes, absolutely. If you take a look at what we've done is we've developed the core technology building blocks, if you will, to serve our primary hyperscale customers. Now we're seeing the second wave of customers come in, the so-called Neo cloud, CSPs or even enterprise customers who are looking now for more rack level solutions. We can repurpose those technology blocks to their needs with just a little bit of customization. And so we get good leverage, we think, on the engineering investment that we've made. We don't have to create all new solutions. And we can bring very good products to market that are proven, reliable and leading edge to this next wave of customers.

Got it. Got it. Now I'm going to move over to semi, which I'm probably a little bit more comfortable with. So you've had a nice year in semi. Q4 guidance implies about low teens growth for the full year. Can you talk about what kind of -- what's driven that? And your customers have talked about a really big -- well, not really big, but a good second half 2026 ramp. Just what are you doing in terms of preparing for what the customers have kind of talked about publicly?

Yes. I think the real story in semiconductor is our success with the new products because that's laying the groundwork for sustained share gain moving forward. So in '23, we introduced 2 new platforms called eVerest and eVoS together with an accompanying product called NavX. And we've had great success on the conductor etch side of the business. So you're going to start seeing that in the form of significant revenue next year. We're also having success in the dielectric etch part of the business, and that's a new area for us actually. And so we anticipate significant revenue starting in 2027 in that area. So we're expanding our TAM, and we're going to grow our market share over the next 5 years based on the success of our new products. I think short term, all of our customers are saying similar things about second half strengthening next year and 2027 being a really good year. And I think that's based on a number of factors, including when new fabs are coming online.

Got it. Yes. So as we kind of think about we're headed hopefully towards a pretty good '26 and '27 for WFE, you guys have laid out a goal where I think it was 20% or 1.2x or 1.3x above WFE. How much of that can we kind of attribute to these new products? And how much of that do you think we should think about in terms of just etch dep intensity increasing within WFE?

Yes, it's a little bit of both. I think our customers have been vocal about the fact that etch and depth intensity is going up. And that's the nature of these new processes, just a lot more steps, a lot more etch and deposition going on. So that's good. It's good for our customers. It's good for us. In addition, we think we get a bump from our success with the new products. So you're going to see eVerest and eVoS in many of these new processes that are ramping over the next 2 or 3 years. And that's both in logic and memory applications.

Got it. And I guess between logic and memory, how would you sort of characterize your exposure? Would you characterize -- how would you characterize sort of your excitement levels for the 2 end markets into '26?

I think at the leading edge, it's pretty exciting across the board. I think certainly for leading-edge logic, the AI demand is driving these new process developments. It's driving volume. And obviously, the leader there is TSMC, and they're doing quite well. On the memory side, very exciting as well with high-bandwidth memory, DRAM and NAND. Also, I think all 3 of those markets will be strong next year, probably into '27. So I think we've turned the corner on leading edge in semiconductor. We're pretty excited about our ability to grow share in all these segments.

And I want to go back to eVerest and eVoS. But as sort of your main customers come over to you, what are they -- how much -- how demanding are these customers in terms of asking you for R&D, getting you to sort of kind of enable the next wave of leading-edge semiconductors?

Yes. I think semiconductor customers are among the most demanding, right? They're very focused on developing solutions that meet the needs of their customers, the fab operators. And so it's an intense process. I said we launched the platforms back in '23. It typically takes 2, maybe 2.5 years to customize the product, and it's a 3-party development. It's us, it's our customers and it's their customers. And so we basically tune our subsystems to their chambers and then work with them to tune it further to the end customer processes.

Got it. And just last, a little bit more shorter-term question. How would you characterize sort of the inventory situation at your customers? Or just given the customer concentration for semis, is that not something we probably need to think about too much?

Yes. I think the inventory situation is healthy. I wouldn't say there's too little or too much. It's about where it should be. So we deal with our major semiconductor customers on a JIT basis. And so we basically fill the bins to wherever they need them fill to. And that's going to vary depending on their view of the market, but we're keeping up with demand. And we have upside capacity at our factory in Malaysia. And ultimately, we're going to qualify the semi products in both Malaysia and Thailand. We have 2 factories.

And you're speaking with these customers every week, so you're on top of it.

Yes.

Got it. So I want to move over to industrial and medical. And I guess, to be a little bit critical, data center has absolutely exploded. Industrial, medical has been a little bit lackluster. But could you talk about what you see in that what you're seeing in that end market, what you're seeing in that business segment and how you're kind of thinking about the 2026 setup for here?

Yes. I think Industrial Medical has been in a correction mode for almost 2 years. And so I think the industrial medical customers, many of them suffered with the COVID supply chain shortages, and it's taken them a while to work through their inventories. I think today, most customers have worked through the inventory. The rest of them should be finished in the next 3 to 6 months. So we're seeing incrementally improving business in '25. We hit bottom in Q1 of '25. And each quarter since we've been a little bit higher in that business, a little higher revenue. We see some good trends in distribution, where the inventories have declined 7 quarters in a row, and we're at equilibrium now in distribution. So we think in Industrial Medical, '26 will be a growth year. It's not going to be a V-shaped recovery. We think it's going to be gradual. But we have a very solid design win pipeline, and we think we could start showing outsized revenue growth in '26 and '27.

I guess to ask a deeper question. So I feel like in semiconductors, when we're in an up cycle, we call it secular. When we're in a down cycle, we call semiconductor cyclical. But just how does this industrial and medical portfolio play a part within the Advanced Energy portfolio, just given how cyclical in hindsight, semiconductors and data center may be?

Yes. I think industrial and medical plays an important part. So the way we think about the business, we have 3 pillars. We have semiconductor, we have data center, we have industrial medical. And at any one point in time, at least one of those markets will be up. And so we can sustain our R&D spending, our capacity investments because under any condition, we're generating good cash flow as a company. So we think it strengthens us as a company, enables us to create some space between us and our competitors. That's the overall view. Industrial Medical is the weakest of 3 pillars. So part of that's market-based. Part of it is our starting position. So moving forward, our primary focus for M&A will be adding more to our I&M portfolio. In addition, we've been investing for the last 3 years in strengthening our channel and strengthening our direct sales force calling on industrial medical accounts. So we think the organic efforts, together with some inorganic efforts will boost us to a #1 position in industrial medical.

Got it. I want to move over to some financial questions, and then I'll open it up to the audience. But -- so you're on track to about, I think, low to mid- $6 of EPS this year, in line with your sort of $1.75 billion revenue model. Assuming we get a good year in semi and data center, it doesn't feel like we're too far away from you guys' $2.5 billion target you laid out for 2030, but it feels like it's not too far away. Just not asking you to update your guidance, but just relative to when you kind of put out this 2030 model, how different is the operating environment been for you guys?

Yes. I think the biggest change that we've talked about it is really the explosion of the data center market. And that's really come on seeing, as you've seen, all of AI explode, combined with our change of strategy Steve mentioned earlier about targeting specific areas which we've been able to win. So that market is running well ahead of what we thought. And in fact, we'll probably achieve our 2030 target that we set a year ago in data center either this year or early next year. So we're well on track there, a little bit ahead. And as we said, as we look forward, we think that's sustainable at these levels or higher, at least over the next year or so with reason to believe with the second wave of customers that we can sustain that and even grow that beyond that point. Semiconductor is -- this 2025 is a good year for us. We'll grow, and it will be our second best semiconductor year ever. Having said that, if you go back a year ago, the WFE projections for growth sound a lot like they sound now. I think the difference is now I think there's a lot more end market data points to corroborate the 2026 second half, we'll see growth. So I think we're generally on track there if you looked at a post '26, early '27 level. Industrial and Medical, as you pointed out, frankly, was a little softer. Our trough level in Q1 of this year was lower than we expected, but we're slowly building out of it. So we feel confident in the model that we are on track from a revenue perspective to achieve $3 billion, and we could achieve it early. I think it's very possible depending on how the markets go. I'll also just add that we feel like we're well on track from a gross margin perspective.

I guess I want to touch on that margin portion because I guess the next step for you guys will be the $2.25 billion level, which is 30% growth from here. But when I think about that, your growth -- gross profit margin should be 42%, operating profit margin, 20%. I feel like you've also been spending a lot more on CapEx just given where end demand is. Can you kind of talk about how that sort of increased CapEx or sort of a changing end market mix per se may sort of impact that financial model that you are envisioning?

Yes, it's a really good question. When we look at gross margins, we have improved them almost 400 basis points over the last 5 quarters and 280 basis points just last quarter alone year-over-year. So we think we're on track. We delivered about half the gross margin improvement to get to the 43% that we talked about earlier. From a CapEx perspective and the additional headwind that goes with that, I'm not too worried about that because that CapEx was contemplated in our model. Bringing on Thailand was always contemplated. It was contemplated at certain revenue levels. We're achieving those revenue levels earlier. So I think that works well within the model. Frankly, what wasn't contemplated was the tariff environment, which has been about 100 basis points of headwind for us. And what wasn't contemplated was the high level of mix of data center at this point. Now semi will recover, and I think that mix will normalize a little bit over the next couple of years. Having said that, we believe we can still achieve the 43%. There's other levers we can pull operationally to help offset the tariff impact if those don't get mitigated hopefully on their own. And we see opportunities, frankly, to continue to improve overall margins and absorb the data center mix. Frankly, we think it's pretty positive that we've absorbed a roughly doubling of our data center business, and we're still on track to get close to 40% exiting this year or early next year.

Got it. So can I -- so when I think about the 3 drivers of gross margin improvement, which is manufacturing and cost improvements, just product mix and then volume leverage, would you say that kind of the contribution from them has been a little bit more skewed towards the volume leverage over the last year or so?

Well, I think the first thing is we've executed on our factory consolidation plan. So we talked about 200 to 250 basis points of improvement on that. We've largely achieved that, not completely, but largely. And like I said, I think we can actually do more there. So over time, we'll get a little bit more. We've been right on track with volume. So certainly, volumes up. That's fit right with our model of how we would expect that to occur. The third leg of margin improvement that we haven't realized as much yet is the portfolio mix. And portfolio mix has mostly to do with as our new products take hold in the market and become a bigger portion of our revenue, those come with higher margins. And we think there's 200 to 300 basis points of improvement that relate to that. Now that won't happen immediately because that products have to get in the market, they have to ramp to scale. But we feel very good about that aspect that, that will occur over time. So on balance, we feel good about the strategy and the execution that we've had and think we'll still be able to get to the 43% ultimately and overcome some of the headwinds we have.

Got it. And just one more question for me, and then I'll pass it over to the audience. But just I&M and sort of how M&A sort of kind of plays into the growth strategy there?

Yes. So if you look at I&M, Industrial Medical, it's a highly fragmented market. So we think there's an opportunity to do a partial roll-up. We play in the high end of the I&M market. Let's say, the top 1/3 of the market is where we play. So we're looking for similar companies that are out there who provide these industrial medical products, which are sticky, right? Once they're designed in, they stay designed in for a very long time. They tend to be long life cycle, 10, 15, 20 years, and they tend to be very profitable. So that's our kind of business. It's hard to displace existing competitors. So you have to go out and do some M&A to really increase market share over the short to medium term.

With that, I'd like to open up to the audience if there's any questions. If not, I'll have plenty of questions about semiconductors, but -- Damn. Actually, I want to ask about data center there. So this 25% to 30% growth for next year, how would you sort of characterize the risk to the upside, just given 100% was probably maybe a little bit higher than what you guys are envisioning this year? Just how would you characterize that risk to the upside?

So risk to the upside, does that mean is there an upside?

Yes.

I think there are opportunities. We charted out '26 based on the design wins we have with our Tier 1 customers. We have not included Tier 2 customer opportunities in '26. We have those penciled in for '27 revenue. But is there a potential to pull it into '26? Yes. but we're not at that point where we could confirm it's going to happen. So I think the bias is probably towards the upside in '26 for data center.

But yes, would that be an Advanced Energy capacity issue or Advanced Energy is being a little bit more selective on, okay, these are the projects that we'd want to go after and kind of prioritize?

Yes, definitely not a capacity issue. We've invested a lot. So we have sufficient capacity in the Philippines and in Mexicali today. We can also ramp up Thailand next year. And so that factory is completed. And for about 6 months from a go signal, we could start to produce product in that factory. So we've got substantial upside, I think, in the second half of next year if we need it.

Got it. And...

What's the risk of double ordering.

So just to repeat, the risk of double ordering in data center and data center?

Yes.

Yes. The risk of double ordering seems to be minimal. In fact, as a company, we really haven't experienced double ordering, I think, over the last 5 years. Most of our products are sole source. And so customers order them when they need the products. And so whether it's data center, semiconductor, industrial, medical, we haven't seen double ordering. So it's not a risk for us that I'm aware of.

What's the lifetime of your products, especially in data center?

Yes. So the question is what's the lifetime of our products, particularly in data center. The lifetime is quite long. We build highly reliable products. So I would say, typically, they're going to last for 7 to 10 years. So it depends on the price of the product. Ultimately, the customer can choose to replace our product with a new box. But in the semiconductor market, the pricing is so high that we have a pretty thriving service business where periodically, we bring the boxes back into our shop and calibrate them, repair them, upgrade them. And it's a very good business for the company. But that's primarily semiconductor.

How much part of semi business is the service part?

Service overall is about 10% of total revenues. And so within semiconductor, you could think of it in the mid- to high teens as a percentage of revenue. So it's healthy, and it's been growing and a lot less volatile. Much more stable, yes, absolutely.

I guess going back to that sort of data center product lifetime question. But I would say your kind of customers are constantly asking for new products and innovation there.

Yes. So basically, it's a continuous process. We're innovating every year. So the customers are trying to keep pace with the introduction of new GPUs and equivalents. And so they bring us in early in the design process so that our power solution is ready and it's an integral part of their rack design.

So the hyperscalers are demanding customers?

Yes. So we have a lot of demanding customers, semiconductor and hyperscalers top the list.

Got it. Still have 30 seconds left, but there are no questions, I think we can wrap that up for today -- Oh, sorry.

I mean I can say that the next 2 years, data center could be strong, but I mean, no one expects the amount of CapEx to be the same in 5 years' time, right?

Yes.

So I guess where is the big change to kind of make up for that maybe drop at some point? Or I mean, how is that going to -- to think got the biggest recovery sort of cycle there?

Yes, that goes back to our business strategy. So we realize that semiconductor and data center are cyclical businesses. So at any given point in time, we think though industrial, medical, semiconductor or data center will be up. And so all these businesses go through ups and downs. But because we're diversified and we play in the high end in multiple markets, we think it makes it much easier for us to thrive during the cyclical downturns of these various markets. We've kind of proven that out over the last few years because '23 was a down year in semiconductor. We continue to increase our R&D spending and capacity spending. '24, we had some issues in some markets. And so every year, we've had the ability to continue to increase R&D spending and increase capacity spending despite downturns in certain of our markets.

Great. That brings us to time. Thank you, Steve. Thank you, Paul.

Thanks a lot.

Thank you, Shane. Thanks, everybody.